Introducing Experiments: Creative Tests That Launch, Judge, and Reallocate Themselves

Everyone agrees you should be testing creative constantly. Almost nobody does it well, because doing it right is a grind: launch the variants, remember to check them, resist killing a winner too early or letting a loser bleed budget, and actually conclude with a decision instead of a shrug. It's not hard because it's complicated. It's hard because it's relentless. So we built an engine that does the relentless part.

You approve a single card. From there the engine launches the variants on Meta and Google (ad rotation and split-campaign), evaluates them on a schedule instead of whenever you remember to look, shifts budget to the winners automatically as results come in, and concludes with a verdict — a real decision, not a dashboard you still have to interpret. You get a wizard to set it up, a live scoreboard to watch it run, and a monthly test-budget meter.

"Runs itself" is the part that makes a careful marketer nervous, so the guardrails aren't an afterthought — they're the point. Every experiment runs under a strict budget cap and a kill switch, and nothing starts until you approve that first card. Autonomy without guardrails is just gambling. Autonomy with a cap and a kill switch is leverage. Experiments is rolling out on Growth plans, with Meta and Google wired end-to-end and Microsoft Ads on the way.